So far, we’ve run a bunch of analyses on the data split two different ways - by the WM capacity scores and by the high load DFR performance. Below is a table of the summary of analyses:

Measure	WM	DFR
DFR delay LE	inverted U	linearly increases
constructs	linearly increases	linearly increases
clinical	linearly decreases	linearly decreases
cue period full mask	asymptotic med,high > low in LE	linearly increases for L3 and LE
delay period full mask	inverted U L3 and LE	linearly increases L3 and LE
delay period indiv ROIs	inverted U in LE	linearly increases in LE
probe period full mask	asymptotic med,high>low in LE	linearly increases in LE
FFA	linearly increases in L cue LE	linearly increases in L cue LE
HPC Posterior	inverted U @ L3 L cue, delay, probe	no effect
RSFC within network	no effect	FPCN: med > high VAN: low > high
RSFC across network	inverted U FPCN/CO	no effects
Beta Series connectivity cue	no effect	linearly increases L3: FPCN/FFA, FPCN/HPC, HPC/FFA LE: FPCN/FPCN, FPCN/FFA
Beta Series connectivity delay	LE: inverted U in HPC/FFA	linearly decreases L3: FPCN/HPC
BCT measures	mean participation coefficient: U shaped DMN: U shaped VAN: linear increase	no effect

library(dplyr)

## 
## Attaching package: 'dplyr'

## The following objects are masked from 'package:stats':
## 
##     filter, lag

## The following objects are masked from 'package:base':
## 
##     intersect, setdiff, setequal, union

library(ggplot2)
library(reshape2)
library(psych)

## 
## Attaching package: 'psych'

## The following objects are masked from 'package:ggplot2':
## 
##     %+%, alpha

library(patchwork)
library(R.matlab)

## R.matlab v3.6.2 (2018-09-26) successfully loaded. See ?R.matlab for help.

## 
## Attaching package: 'R.matlab'

## The following objects are masked from 'package:base':
## 
##     getOption, isOpen

load('data/load_effects_DFR.RData')
load('data/behav.RData')
load('data/structural_measures.RData')
load('data/connectivity_data.RData')
load('data/DFR_split_groups_info.RData')
load('data/split_WM_groups_fMRI.RData')

source('load_in_ROI.R')
source('split_TC_into_groups.R')
source('create_TC_for_plot.R')

# these times are based on when the actual cues were on the screen 
rects <- data.frame(xstart=c(0,2.5,10),xend=c(2.5,10,12),col = factor(c("cue","delay","probe"),levels=c("cue","delay","probe")))

# adjust for hemodynamic delay 
rects$xstart <- rects$xstart+5
rects$xend <- rects$xend+5

Regressions

Interestingly, it looks like there is ovearchingly a linear relationship between these measures and DFR performance, with an inverted U-shape relationship with capacity. Let’s see if these relationships hold up when we include both of them in a model, especially for the ones where there is a significant relationship with both variables. For the WM capacity, we’re going to add in a quadratic term to see if that fits the data.

We’re also going to regress the effect of DFR accuracy out of omnibus span and see what the residuals look like plotted against our variables of interest.

base_data <- merge(p200_data,constructs_fMRI,by="PTID")
# want to create a base data that we can include in models - DFR accuracy, WM capacity and gender, age and WHODAS scores as covariates 
base_data <- dplyr::select(base_data,PTID,"XDFR_MRI_ACC_L3","omnibus_span_no_DFR_MRI","X010701_GENDER","PX010101_AGE","WHO_ST_S32")
colnames(base_data) <- c("PTID","DFR_L3_ACC","omnibus_span", "gender", "age", "WHODAS")
base_data$span_sq <- base_data$omnibus_span^2
base_data <- merge(base_data,p200_demographics[,c(1,4)])

# shift to dummy coding 
base_data$SCANNER <- base_data$SCANNER - 1
base_data$gender <- base_data$gender - 1 

# one subject did not report gender - assign 0.5 as to not mess with the regression too much
base_data$gender[73] <- 0.5

# remove subjects with incomplete data 
base_data <- base_data[c(1,90,92:105,107:170),]

For findings where DFR is significant, can try to regess it out of omnibus and plot that against variable of interest to see if inverted U shape is still present, since I’m not sure span^2 is the right measure.

omnibus.lm <- lm(omnibus_span ~ DFR_L3_ACC + gender + age + WHODAS + SCANNER, data= base_data)
summary(omnibus.lm)

## 
## Call:
## lm(formula = omnibus_span ~ DFR_L3_ACC + gender + age + WHODAS + 
##     SCANNER, data = base_data)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -1.31306 -0.24803  0.02675  0.28368  0.79189 
## 
## Coefficients:
##               Estimate Std. Error t value Pr(>|t|)  
## (Intercept) -0.4546861  0.5106225  -0.890   0.3761  
## DFR_L3_ACC   1.3803242  0.5653984   2.441   0.0170 *
## gender      -0.1945473  0.1103004  -1.764   0.0819 .
## age         -0.0109993  0.0104701  -1.051   0.2969  
## WHODAS       0.0002215  0.0043189   0.051   0.9592  
## SCANNER     -0.0837151  0.1205155  -0.695   0.4895  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.4569 on 74 degrees of freedom
## Multiple R-squared:  0.1074, Adjusted R-squared:  0.0471 
## F-statistic: 1.781 on 5 and 74 DF,  p-value: 0.1272

base_data$omnibus_resid <- resid(omnibus.lm)

Cue Period Load Effect

full_mask_data <- merge(base_data,p200_DFR_full_mask)

cue_loadEffect_DFR.lm <- lm(cue_loadEffect ~ DFR_L3_ACC + gender + age + WHODAS + SCANNER, data= full_mask_data)
summary(cue_loadEffect_DFR.lm)

## 
## Call:
## lm(formula = cue_loadEffect ~ DFR_L3_ACC + gender + age + WHODAS + 
##     SCANNER, data = full_mask_data)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -0.90531 -0.22948 -0.00748  0.17708  0.67657 
## 
## Coefficients:
##               Estimate Std. Error t value Pr(>|t|)    
## (Intercept) -0.4136513  0.3477157  -1.190 0.237997    
## DFR_L3_ACC   1.3941795  0.3850162   3.621 0.000534 ***
## gender      -0.0236037  0.0751106  -0.314 0.754213    
## age         -0.0070080  0.0071297  -0.983 0.328847    
## WHODAS       0.0006013  0.0029410   0.204 0.838560    
## SCANNER     -0.1337834  0.0820667  -1.630 0.107313    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.3112 on 74 degrees of freedom
## Multiple R-squared:  0.2007, Adjusted R-squared:  0.1467 
## F-statistic: 3.716 on 5 and 74 DF,  p-value: 0.004687

cue_loadEffect_span.lm <- lm(cue_loadEffect ~ omnibus_span + span_sq + gender + age + WHODAS + SCANNER, data= full_mask_data)
summary(cue_loadEffect_span.lm)

## 
## Call:
## lm(formula = cue_loadEffect ~ omnibus_span + span_sq + gender + 
##     age + WHODAS + SCANNER, data = full_mask_data)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -0.92986 -0.20437 -0.00479  0.22024  0.66084 
## 
## Coefficients:
##                Estimate Std. Error t value Pr(>|t|)  
## (Intercept)   0.5153954  0.2322313   2.219   0.0296 *
## omnibus_span  0.1040112  0.0826575   1.258   0.2123  
## span_sq       0.1044008  0.1226151   0.851   0.3973  
## gender        0.0544244  0.0789557   0.689   0.4928  
## age          -0.0056419  0.0077184  -0.731   0.4671  
## WHODAS        0.0001114  0.0031597   0.035   0.9720  
## SCANNER      -0.1364229  0.0884565  -1.542   0.1273  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.3338 on 73 degrees of freedom
## Multiple R-squared:  0.09258,    Adjusted R-squared:  0.018 
## F-statistic: 1.241 on 6 and 73 DF,  p-value: 0.2955

cue_loadEffect_all.lm <- lm(cue_loadEffect ~ omnibus_span + span_sq + DFR_L3_ACC + gender + age + WHODAS + SCANNER, data = full_mask_data)
summary(cue_loadEffect_all.lm)

## 
## Call:
## lm(formula = cue_loadEffect ~ omnibus_span + span_sq + DFR_L3_ACC + 
##     gender + age + WHODAS + SCANNER, data = full_mask_data)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -0.89065 -0.22578 -0.01418  0.16702  0.66804 
## 
## Coefficients:
##                Estimate Std. Error t value Pr(>|t|)   
## (Intercept)  -0.3799097  0.3547926  -1.071  0.28784   
## omnibus_span  0.0388032  0.0804741   0.482  0.63114   
## span_sq       0.0505711  0.1167131   0.433  0.66610   
## DFR_L3_ACC    1.3098717  0.4086722   3.205  0.00201 **
## gender       -0.0180476  0.0777342  -0.232  0.81706   
## age          -0.0067786  0.0072790  -0.931  0.35484   
## WHODAS        0.0006976  0.0029819   0.234  0.81568   
## SCANNER      -0.1279304  0.0833645  -1.535  0.12927   
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.3144 on 72 degrees of freedom
## Multiple R-squared:  0.2059, Adjusted R-squared:  0.1287 
## F-statistic: 2.667 on 7 and 72 DF,  p-value: 0.01641

no_resid <- ggplot(data = full_mask_data, aes(x=omnibus_span,y=cue_loadEffect))+
  geom_point()+
  stat_smooth(method="loess")+
  ggtitle("DFR not regressed")

resid <- ggplot(data = full_mask_data, aes(x=omnibus_resid,y=cue_loadEffect))+
  geom_point()+
  stat_smooth(method="loess")+
  ggtitle("DFR regressed")

no_resid + resid+
  plot_annotation(title = "Cue Load Effect")

cor.test(full_mask_data$omnibus_resid,full_mask_data$cue_loadEffect)

## 
##  Pearson's product-moment correlation
## 
## data:  full_mask_data$omnibus_resid and full_mask_data$cue_loadEffect
## t = 0.49495, df = 78, p-value = 0.622
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
##  -0.1658010  0.2723232
## sample estimates:
##        cor 
## 0.05595424

Delay Period

Interestingly, the delay period is the only finding that had an inverted U-shape relationship with capacity that held up when we regressed out the effects of accuracy.

High Load

delay_high_DFR.lm <- lm(delay_high ~ DFR_L3_ACC + gender + age + WHODAS + SCANNER, data= full_mask_data)
summary(delay_high_DFR.lm)

## 
## Call:
## lm(formula = delay_high ~ DFR_L3_ACC + gender + age + WHODAS + 
##     SCANNER, data = full_mask_data)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -0.38378 -0.13449 -0.03422  0.09021  0.62031 
## 
## Coefficients:
##              Estimate Std. Error t value Pr(>|t|)   
## (Intercept) -0.546914   0.242262  -2.258  0.02692 * 
## DFR_L3_ACC   0.737987   0.268250   2.751  0.00746 **
## gender       0.045753   0.052331   0.874  0.38478   
## age          0.001946   0.004967   0.392  0.69632   
## WHODAS       0.001044   0.002049   0.509  0.61194   
## SCANNER     -0.047811   0.057178  -0.836  0.40575   
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.2168 on 74 degrees of freedom
## Multiple R-squared:  0.144,  Adjusted R-squared:  0.08618 
## F-statistic:  2.49 on 5 and 74 DF,  p-value: 0.03859

delay_high_span.lm <- lm(delay_high ~ omnibus_span + span_sq + gender + age + WHODAS + SCANNER, data= full_mask_data)
summary(delay_high_span.lm)

## 
## Call:
## lm(formula = delay_high ~ omnibus_span + span_sq + gender + age + 
##     WHODAS + SCANNER, data = full_mask_data)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -0.40421 -0.14809 -0.03454  0.10732  0.59040 
## 
## Coefficients:
##               Estimate Std. Error t value Pr(>|t|)
## (Intercept)  -0.025694   0.159322  -0.161    0.872
## omnibus_span  0.015558   0.056707   0.274    0.785
## span_sq      -0.017290   0.084120  -0.206    0.838
## gender        0.087375   0.054168   1.613    0.111
## age           0.002570   0.005295   0.485    0.629
## WHODAS        0.000586   0.002168   0.270    0.788
## SCANNER      -0.057641   0.060686  -0.950    0.345
## 
## Residual standard error: 0.229 on 73 degrees of freedom
## Multiple R-squared:  0.05779,    Adjusted R-squared:  -0.01965 
## F-statistic: 0.7463 on 6 and 73 DF,  p-value: 0.6143

delay_high_all.lm <- lm(delay_high ~ omnibus_span + span_sq + DFR_L3_ACC + gender + age + WHODAS + SCANNER, data = full_mask_data)
summary(delay_high_all.lm)

## 
## Call:
## lm(formula = delay_high ~ omnibus_span + span_sq + DFR_L3_ACC + 
##     gender + age + WHODAS + SCANNER, data = full_mask_data)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -0.39929 -0.12964 -0.03817  0.08507  0.61096 
## 
## Coefficients:
##               Estimate Std. Error t value Pr(>|t|)   
## (Intercept)  -0.573996   0.246907  -2.325  0.02291 * 
## omnibus_span -0.024377   0.056003  -0.435  0.66467   
## span_sq      -0.050256   0.081223  -0.619  0.53804   
## DFR_L3_ACC    0.802191   0.284403   2.821  0.00619 **
## gender        0.042991   0.054097   0.795  0.42939   
## age           0.001874   0.005066   0.370  0.71246   
## WHODAS        0.000945   0.002075   0.455  0.65019   
## SCANNER      -0.052440   0.058015  -0.904  0.36906   
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.2188 on 72 degrees of freedom
## Multiple R-squared:  0.1515, Adjusted R-squared:  0.06906 
## F-statistic: 1.837 on 7 and 72 DF,  p-value: 0.09314

no_resid <- ggplot(data = full_mask_data, aes(x=omnibus_span,y=delay_high))+
  geom_point()+
  stat_smooth(method="loess")+
  ggtitle("DFR not regressed")

resid <- ggplot(data = full_mask_data, aes(x=omnibus_resid,y=delay_high))+
  geom_point()+
  stat_smooth(method="loess")+
  ggtitle("DFR regressed")

no_resid + resid+
  plot_annotation(title = "Delay High Load")

cor.test(full_mask_data$omnibus_resid,full_mask_data$delay_high)

## 
##  Pearson's product-moment correlation
## 
## data:  full_mask_data$omnibus_resid and full_mask_data$delay_high
## t = -0.48586, df = 78, p-value = 0.6284
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
##  -0.2713718  0.1667999
## sample estimates:
##         cor 
## -0.05493016

Load Effect

delay_loadEffect_DFR.lm <- lm(delay_loadEffect ~ DFR_L3_ACC + gender + age + WHODAS + SCANNER, data= full_mask_data)
summary(delay_loadEffect_DFR.lm)

## 
## Call:
## lm(formula = delay_loadEffect ~ DFR_L3_ACC + gender + age + WHODAS + 
##     SCANNER, data = full_mask_data)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -0.59393 -0.11461 -0.02709  0.15460  0.63192 
## 
## Coefficients:
##              Estimate Std. Error t value Pr(>|t|)   
## (Intercept) -0.384499   0.249247  -1.543  0.12718   
## DFR_L3_ACC   0.819120   0.275985   2.968  0.00404 **
## gender      -0.070746   0.053840  -1.314  0.19290   
## age         -0.003607   0.005111  -0.706  0.48251   
## WHODAS       0.004785   0.002108   2.270  0.02615 * 
## SCANNER     -0.027070   0.058827  -0.460  0.64675   
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.223 on 74 degrees of freedom
## Multiple R-squared:  0.1759, Adjusted R-squared:  0.1202 
## F-statistic: 3.158 on 5 and 74 DF,  p-value: 0.01223

delay_loadEffect_span.lm <- lm(delay_loadEffect ~ omnibus_span + span_sq + gender + age + WHODAS + SCANNER, data= full_mask_data)
summary(delay_loadEffect_span.lm)

## 
## Call:
## lm(formula = delay_loadEffect ~ omnibus_span + span_sq + gender + 
##     age + WHODAS + SCANNER, data = full_mask_data)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -0.59973 -0.14499  0.00264  0.14608  0.70624 
## 
## Coefficients:
##               Estimate Std. Error t value Pr(>|t|)  
## (Intercept)   0.182020   0.164718   1.105   0.2728  
## omnibus_span  0.027008   0.058628   0.461   0.6464  
## span_sq       0.039507   0.086969   0.454   0.6510  
## gender       -0.027530   0.056002  -0.492   0.6245  
## age          -0.003063   0.005475  -0.559   0.5776  
## WHODAS        0.004424   0.002241   1.974   0.0522 .
## SCANNER      -0.033450   0.062741  -0.533   0.5956  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.2368 on 73 degrees of freedom
## Multiple R-squared:  0.08396,    Adjusted R-squared:  0.008668 
## F-statistic: 1.115 on 6 and 73 DF,  p-value: 0.362

delay_loadEffect_all.lm <- lm(delay_loadEffect ~ omnibus_span + span_sq + DFR_L3_ACC + gender + age + WHODAS + SCANNER, data = full_mask_data)
summary(delay_loadEffect_all.lm)

## 
## Call:
## lm(formula = delay_loadEffect ~ omnibus_span + span_sq + DFR_L3_ACC + 
##     gender + age + WHODAS + SCANNER, data = full_mask_data)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -0.58744 -0.12372 -0.02423  0.13848  0.63198 
## 
## Coefficients:
##               Estimate Std. Error t value Pr(>|t|)   
## (Intercept)  -0.389508   0.255036  -1.527  0.13108   
## omnibus_span -0.014618   0.057847  -0.253  0.80122   
## span_sq       0.005144   0.083897   0.061  0.95128   
## DFR_L3_ACC    0.836170   0.293767   2.846  0.00576 **
## gender       -0.073793   0.055878  -1.321  0.19081   
## age          -0.003788   0.005232  -0.724  0.47142   
## WHODAS        0.004798   0.002143   2.239  0.02827 * 
## SCANNER      -0.028028   0.059925  -0.468  0.64139   
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.226 on 72 degrees of freedom
## Multiple R-squared:  0.1766, Adjusted R-squared:  0.09656 
## F-statistic: 2.206 on 7 and 72 DF,  p-value: 0.04351

no_resid <- ggplot(data = full_mask_data, aes(x=omnibus_span,y=delay_loadEffect))+
  geom_point()+
  stat_smooth(method="loess")+
  ggtitle("DFR not regressed")

resid <- ggplot(data = full_mask_data, aes(x=omnibus_resid,y=delay_loadEffect))+
  geom_point()+
  stat_smooth(method="loess")+
  ggtitle("DFR regressed")

no_resid + resid+
  plot_annotation(title = "Delay Load Effect")

cor.test(full_mask_data$omnibus_resid,full_mask_data$delay_loadEffect)

## 
##  Pearson's product-moment correlation
## 
## data:  full_mask_data$omnibus_resid and full_mask_data$delay_loadEffect
## t = -0.23381, df = 78, p-value = 0.8157
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
##  -0.2447586  0.1943823
## sample estimates:
##         cor 
## -0.02646493

Probe Period LE

probe_loadEffect_DFR.lm <- lm(probe_loadEffect ~ DFR_L3_ACC + gender + age + WHODAS + SCANNER, data= full_mask_data)
summary(probe_loadEffect_DFR.lm)

## 
## Call:
## lm(formula = probe_loadEffect ~ DFR_L3_ACC + gender + age + WHODAS + 
##     SCANNER, data = full_mask_data)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -1.88480 -0.45535  0.00185  0.46013  1.69318 
## 
## Coefficients:
##               Estimate Std. Error t value Pr(>|t|)  
## (Intercept) -0.4399094  0.7900589  -0.557   0.5793  
## DFR_L3_ACC   1.7414872  0.8748107   1.991   0.0502 .
## gender      -0.0485565  0.1706618  -0.285   0.7768  
## age         -0.0150524  0.0161998  -0.929   0.3558  
## WHODAS       0.0008446  0.0066823   0.126   0.8998  
## SCANNER      0.1276699  0.1864671   0.685   0.4957  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.707 on 74 degrees of freedom
## Multiple R-squared:  0.06473,    Adjusted R-squared:  0.001533 
## F-statistic: 1.024 on 5 and 74 DF,  p-value: 0.4097

probe_loadEffect_span.lm <- lm(probe_loadEffect ~ omnibus_span + span_sq + gender + age + WHODAS + SCANNER, data= full_mask_data)
summary(probe_loadEffect_span.lm)

## 
## Call:
## lm(formula = probe_loadEffect ~ omnibus_span + span_sq + gender + 
##     age + WHODAS + SCANNER, data = full_mask_data)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -1.8839 -0.4501  0.1026  0.4875  1.9026 
## 
## Coefficients:
##                Estimate Std. Error t value Pr(>|t|)
## (Intercept)   7.045e-01  5.045e-01   1.396    0.167
## omnibus_span  1.872e-01  1.796e-01   1.043    0.301
## span_sq       6.769e-03  2.663e-01   0.025    0.980
## gender        6.472e-02  1.715e-01   0.377    0.707
## age          -1.224e-02  1.677e-02  -0.730    0.468
## WHODAS       -3.431e-05  6.864e-03  -0.005    0.996
## SCANNER       1.229e-01  1.921e-01   0.639    0.525
## 
## Residual standard error: 0.7251 on 73 degrees of freedom
## Multiple R-squared:  0.02956,    Adjusted R-squared:  -0.05021 
## F-statistic: 0.3705 on 6 and 73 DF,  p-value: 0.8954

probe_loadEffect_all.lm <- lm(probe_loadEffect ~ omnibus_span + span_sq + DFR_L3_ACC + gender + age + WHODAS + SCANNER, data = full_mask_data)
summary(probe_loadEffect_all.lm)

## 
## Call:
## lm(formula = probe_loadEffect ~ omnibus_span + span_sq + DFR_L3_ACC + 
##     gender + age + WHODAS + SCANNER, data = full_mask_data)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -1.92924 -0.43279  0.01915  0.44137  1.71501 
## 
## Coefficients:
##               Estimate Std. Error t value Pr(>|t|)  
## (Intercept)  -0.410914   0.806746  -0.509   0.6121  
## omnibus_span  0.105981   0.182986   0.579   0.5643  
## span_sq      -0.060293   0.265388  -0.227   0.8209  
## DFR_L3_ACC    1.631856   0.929260   1.756   0.0833 .
## gender       -0.025562   0.176756  -0.145   0.8854  
## age          -0.013651   0.016552  -0.825   0.4122  
## WHODAS        0.000696   0.006780   0.103   0.9185  
## SCANNER       0.133437   0.189558   0.704   0.4837  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.715 on 72 degrees of freedom
## Multiple R-squared:  0.06941,    Adjusted R-squared:  -0.02106 
## F-statistic: 0.7672 on 7 and 72 DF,  p-value: 0.6165

no_resid <- ggplot(data = full_mask_data, aes(x=omnibus_span,y=probe_loadEffect))+
  geom_point()+
  stat_smooth(method="loess")+
  ggtitle("DFR not regressed")

resid <- ggplot(data = full_mask_data, aes(x=omnibus_resid,y=probe_loadEffect))+
  geom_point()+
  stat_smooth(method="loess")+
  ggtitle("DFR regressed")

no_resid + resid+
  plot_annotation(title = "Probe Load Effect")

cor.test(full_mask_data$omnibus_resid,full_mask_data$probe_loadEffect)

## 
##  Pearson's product-moment correlation
## 
## data:  full_mask_data$omnibus_resid and full_mask_data$probe_loadEffect
## t = 0.56106, df = 78, p-value = 0.5764
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
##  -0.1585256  0.2792269
## sample estimates:
##        cor 
## 0.06339958

FFA

Looking at the load effect in the L FFA during the cue period.

FFA_data <- merge(base_data,p200_FFA)

FFA_loadEffect_DFR.lm <- lm(L_CUE_LE ~ DFR_L3_ACC + gender + age + WHODAS + SCANNER, data= FFA_data)
summary(FFA_loadEffect_DFR.lm)

## 
## Call:
## lm(formula = L_CUE_LE ~ DFR_L3_ACC + gender + age + WHODAS + 
##     SCANNER, data = FFA_data)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -1.07509 -0.30925 -0.06537  0.30747  1.65380 
## 
## Coefficients:
##              Estimate Std. Error t value Pr(>|t|)   
## (Intercept) -0.911835   0.551217  -1.654  0.10232   
## DFR_L3_ACC   1.972335   0.610347   3.231  0.00184 **
## gender       0.184215   0.119069   1.547  0.12610   
## age          0.001152   0.011302   0.102  0.91911   
## WHODAS      -0.001668   0.004662  -0.358  0.72150   
## SCANNER     -0.123518   0.130096  -0.949  0.34549   
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.4933 on 74 degrees of freedom
## Multiple R-squared:  0.1969, Adjusted R-squared:  0.1426 
## F-statistic: 3.628 on 5 and 74 DF,  p-value: 0.00545

FFA_loadEffect_span.lm <- lm(L_CUE_LE ~ omnibus_span + gender + age + WHODAS + SCANNER, data= FFA_data)
summary(FFA_loadEffect_span.lm)

## 
## Call:
## lm(formula = L_CUE_LE ~ omnibus_span + gender + age + WHODAS + 
##     SCANNER, data = FFA_data)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -1.01990 -0.34760 -0.04012  0.36158  1.56780 
## 
## Coefficients:
##               Estimate Std. Error t value Pr(>|t|)   
## (Intercept)   0.350032   0.354705   0.987  0.32694   
## omnibus_span  0.278021   0.124843   2.227  0.02899 * 
## gender        0.321146   0.119636   2.684  0.00896 **
## age           0.005052   0.011754   0.430  0.66856   
## WHODAS       -0.002644   0.004808  -0.550  0.58399   
## SCANNER      -0.122622   0.134860  -0.909  0.36617   
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.5101 on 74 degrees of freedom
## Multiple R-squared:  0.1411, Adjusted R-squared:  0.08307 
## F-statistic: 2.431 on 5 and 74 DF,  p-value: 0.04266

FFA_loadEffect_all.lm <- lm(L_CUE_LE ~ omnibus_span +  DFR_L3_ACC + gender + age + WHODAS + SCANNER, data = FFA_data)
summary(FFA_loadEffect_all.lm)

## 
## Call:
## lm(formula = L_CUE_LE ~ omnibus_span + DFR_L3_ACC + gender + 
##     age + WHODAS + SCANNER, data = FFA_data)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -1.06117 -0.32496 -0.04474  0.33359  1.58712 
## 
## Coefficients:
##               Estimate Std. Error t value Pr(>|t|)   
## (Intercept)  -0.827569   0.549661  -1.506  0.13648   
## omnibus_span  0.185328   0.124470   1.489  0.14081   
## DFR_L3_ACC    1.716522   0.629297   2.728  0.00798 **
## gender        0.220271   0.120559   1.827  0.07178 . 
## age           0.003190   0.011294   0.282  0.77838   
## WHODAS       -0.001709   0.004624  -0.370  0.71274   
## SCANNER      -0.108003   0.129460  -0.834  0.40686   
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.4893 on 73 degrees of freedom
## Multiple R-squared:  0.2205, Adjusted R-squared:  0.1565 
## F-statistic: 3.443 on 6 and 73 DF,  p-value: 0.004754

no_resid <- ggplot(data = FFA_data, aes(x=omnibus_span,y=L_CUE_LE))+
  geom_point()+
  stat_smooth(method="loess")+
  ggtitle("DFR not regressed")

resid <- ggplot(data = FFA_data, aes(x=omnibus_resid,y=L_CUE_LE))+
  geom_point()+
  stat_smooth(method="loess")+
  ggtitle("DFR regressed")

no_resid + resid+
  plot_annotation(title = "L FFA - cue LE")

cor.test(FFA_data$omnibus_resid,FFA_data$L_CUE_LE)

## 
##  Pearson's product-moment correlation
## 
## data:  FFA_data$omnibus_resid and FFA_data$L_CUE_LE
## t = 1.3752, df = 78, p-value = 0.173
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
##  -0.06816677  0.36135593
## sample estimates:
##       cor 
## 0.1538545

Time courses

HL <- split_constructs[["low"]][split_constructs[["low"]]$PTID %in% DFR_median_split_groups[["high"]]$PTID,]
HL$level <- "HL"
HM <- split_constructs[["med"]][split_constructs[["med"]]$PTID %in% DFR_median_split_groups[["high"]]$PTID,]
HM$level <- "HM"
HH <- split_constructs[["high"]][split_constructs[["high"]]$PTID %in% DFR_median_split_groups[["high"]]$PTID,]
HH$level <- "HH"

LL <- split_constructs[["low"]][split_constructs[["low"]]$PTID %in% DFR_median_split_groups[["low"]]$PTID,]
LL$level <- "LL"
LM <- split_constructs[["med"]][split_constructs[["med"]]$PTID %in% DFR_median_split_groups[["low"]]$PTID,]
LM$level <- "LM"
LH <- split_constructs[["high"]][split_constructs[["high"]]$PTID %in% DFR_median_split_groups[["low"]]$PTID,]
LH$level <- "LH"

things_to_hist <- rbind(HL,HM,HH,LL,LM,LH)

six_groups <- list(HL = data.frame(HL), HM = data.frame(HM), HH = data.frame(HH), LL = data.frame(LL), LM = data.frame(LM), LH = data.frame(LH), all = data.frame(things_to_hist))

basepath <- "~/Documents/UCLA/Research/RDoC/TimeCourseData/"

# delay period 

delay_ROI_list <- c("L_dlPFC", "L_aMFG", "L_dMFG", "L_IPS", "L_preSMA", "R_dlPFC", "R_dMFG",
                    "R_IPS", "R_medParietal")

delay_TCs <- load_in_ROI(basepath, delay_ROI_list)

# cue 

cue_ROI_list <- c("cue_R_preSMA", "cue_R_occipital", "cue_R_MFG", "cue_R_IPS", "cue_R_insula", 
                  "cue_R_FEF", "cue_L_occipital", "cue_L_IPS", "cue_L_insula", "cue_L_FEF")

cue_TCs <- load_in_ROI(basepath, cue_ROI_list)

# probe 

probe_ROI_list <- c("probe_R_OFC", "probe_R_insula", "probe_R_dlPFC", "probe_L_IPS", "probe_L_insula",
                    "probe_L_dlPFC", "probe_L_aMFG", "probe_dmPFC")

probe_TCs <- load_in_ROI(basepath, probe_ROI_list)

allSubjs <- constructs_fMRI$PTID

split_cue_six_groups <- split_TC_into_groups(cue_TCs,six_groups,allSubjs,group_names=c("HL","HM","HH", "LL", "LM", "LH"))
split_delay_six_groups <- split_TC_into_groups(delay_TCs,six_groups,allSubjs,group_names=c("HL","HM","HH", "LL", "LM", "LH"))
split_probe_six_groups <- split_TC_into_groups(probe_TCs,six_groups,allSubjs,group_names=c("HL","HM","HH", "LL", "LM", "LH"))

cue_TC_for_plot <- create_TC_for_plot(split_cue_six_groups)
delay_TC_for_plot <- create_TC_for_plot(split_delay_six_groups)
probe_TC_for_plot <- create_TC_for_plot(split_probe_six_groups)

L1 vs L3 activity

Here, we’re plotting our data split into 6 groups - based on both WM capacity and a median split DFR performance.

Groups are labeled by this split - the first letter (H/L) is the DFR split and the second letter (H/M/L) is the WM split.

Cue

for (ROI in seq.int(1,length(cue_TC_for_plot))){
  print(ggplot(data=cue_TC_for_plot[[ROI]][["wide"]]) + 
          geom_rect(data=rects,aes(xmin=xstart, xmax=xend, ymin = -Inf, ymax=Inf, fill=col), alpha =0.4,show.legend = FALSE)+
          geom_line(aes(x=Time,y=L3,color=level),size=1)+
          geom_line(aes(x=Time,y=L1,color=level),size=1,linetype="dashed")+
          ylab("Mean Activity") + 
          ggtitle(paste0("L3 vs L1 - ",names(cue_TC_for_plot)[ROI]))+
          ylim(c(-.4,.6)))
}

Delay

for (ROI in seq.int(1,length(delay_TC_for_plot))){
  print(ggplot(data=delay_TC_for_plot[[ROI]][["wide"]]) + 
          geom_rect(data=rects,aes(xmin=xstart, xmax=xend, ymin = -Inf, ymax=Inf, fill=col), alpha =0.4,show.legend = FALSE)+
          geom_line(aes(x=Time,y=L3,color=level),size=1)+
          geom_line(aes(x=Time,y=L1,color=level),size=1,linetype="dashed")+
          ylab("Mean Activity") + 
          ggtitle(paste0("L3 vs L1 - ",names(delay_TC_for_plot)[ROI]))+
          ylim(c(-.4,.6)))
}

Probe

for (ROI in seq.int(1,length(probe_TC_for_plot))){
  print(ggplot(data=probe_TC_for_plot[[ROI]][["wide"]]) + 
          geom_rect(data=rects,aes(xmin=xstart, xmax=xend, ymin = -Inf, ymax=Inf, fill=col), alpha =0.4,show.legend = FALSE)+
          geom_line(aes(x=Time,y=L3,color=level),size=1)+
          geom_line(aes(x=Time,y=L1,color=level),size=1,linetype="dashed")+
          ylab("Mean Activity") + 
          ggtitle(paste0("L3 vs L1 - ",names(probe_TC_for_plot)[ROI]))+
          ylim(c(-.4,.6)))
}

Load Effects

Cue

for (ROI in seq.int(1,length(cue_TC_for_plot))){
  print(ggplot(data=cue_TC_for_plot[[ROI]][["wide"]]) +
          geom_rect(data=rects,aes(xmin=xstart, xmax=xend, ymin = -Inf, ymax=Inf), fill="grey", alpha =0.4,show.legend = FALSE)+
          geom_line(aes(x=Time,y=LE,color=level),size=1)+
          geom_ribbon(data=cue_TC_for_plot[[ROI]][["long"]] %>% filter(load == "LE"),aes(x=Time,ymin=SE_min,ymax=SE_max,fill=level),alpha=.2,linetype=2)+
          ylab("Mean Activity") +
          ggtitle(paste0("LE - ",names(cue_TC_for_plot)[ROI]))+
          ylim(c(-.4,.6))
  )
}

Delay

for (ROI in seq.int(1,length(delay_TC_for_plot))){
  print(ggplot(data=delay_TC_for_plot[[ROI]][["wide"]]) +
          geom_rect(data=rects,aes(xmin=xstart, xmax=xend, ymin = -Inf, ymax=Inf), fill="grey", alpha =0.4,show.legend = FALSE)+
          geom_line(aes(x=Time,y=LE,color=level),size=1)+
          geom_ribbon(data=delay_TC_for_plot[[ROI]][["long"]] %>% filter(load == "LE"),aes(x=Time,ymin=SE_min,ymax=SE_max,fill=level),alpha=.2,linetype=2)+
          ylab("Mean Activity") +
          ggtitle(paste0("LE - ",names(delay_TC_for_plot)[ROI]))+
          ylim(c(-.4,.6))
  )
}

Probe

for (ROI in seq.int(1,length(probe_TC_for_plot))){
  print(ggplot(data=probe_TC_for_plot[[ROI]][["wide"]]) +
          geom_rect(data=rects,aes(xmin=xstart, xmax=xend, ymin = -Inf, ymax=Inf), fill="grey", alpha =0.4,show.legend = FALSE)+
          geom_line(aes(x=Time,y=LE,color=level),size=1)+
          geom_ribbon(data=probe_TC_for_plot[[ROI]][["long"]] %>% filter(load == "LE"),aes(x=Time,ymin=SE_min,ymax=SE_max,fill=level),alpha=.2,linetype=2)+
          ylab("Mean Activity") +
          ggtitle(paste0("LE - ",names(probe_TC_for_plot)[ROI]))+
          ylim(c(-.4,.6))
  )
}

Split on DFR

To be able to compare across groups better, let’s split high performing and low performing groups onto different graphs.

Cue

low_col = c("turquoise","blue","violet")
high_col = c("red","gold","green")

for (ROI in seq.int(1,length(cue_TC_for_plot))){
  low <- ggplot(data=cue_TC_for_plot[[ROI]][["wide"]] %>% filter(level == "LL" | level == "LM" | level == "LH")) +
    geom_rect(data=rects,aes(xmin=xstart, xmax=xend, ymin = -Inf, ymax=Inf), fill="grey", alpha =0.4,show.legend = FALSE)+
    geom_line(aes(x=Time,y=LE,color=level),size=1)+
    geom_ribbon(data=cue_TC_for_plot[[ROI]][["long"]] %>% filter(load=="LE") %>% filter(level == "LL"),aes(x=Time,ymin=SE_min,ymax=SE_max),alpha=.2,linetype=2,fill="turquoise")+
    geom_ribbon(data=cue_TC_for_plot[[ROI]][["long"]] %>% filter(load=="LE") %>% filter(level == "LM"),aes(x=Time,ymin=SE_min,ymax=SE_max),alpha=.2,linetype=2,fill="blue")+
    geom_ribbon(data=cue_TC_for_plot[[ROI]][["long"]] %>% filter(load=="LE") %>% filter(level == "LH"),aes(x=Time,ymin=SE_min,ymax=SE_max),alpha=.2,linetype=2,fill="violet")+
    ylab("Mean Activity") +
    ggtitle("Low Performing Subjects")+
    ylim(c(-.2,.4))+
    scale_colour_manual(values=low_col)
  
  high <- ggplot(data=cue_TC_for_plot[[ROI]][["wide"]] %>% filter(level == "HL" | level == "HM" | level == "HH")) +
    geom_rect(data=rects,aes(xmin=xstart, xmax=xend, ymin = -Inf, ymax=Inf), fill="grey", alpha =0.4,show.legend = FALSE)+
    geom_line(aes(x=Time,y=LE,color=level),size=1)+
    geom_ribbon(data=cue_TC_for_plot[[ROI]][["long"]] %>% filter(load=="LE") %>% filter(level == "HL"),aes(x=Time,ymin=SE_min,ymax=SE_max),fill="red",alpha=.2,linetype=2)+
    geom_ribbon(data=cue_TC_for_plot[[ROI]][["long"]] %>% filter(load=="LE") %>% filter(level == "HM"),aes(x=Time,ymin=SE_min,ymax=SE_max),fill="gold",alpha=.2,linetype=2)+
    geom_ribbon(data=cue_TC_for_plot[[ROI]][["long"]] %>% filter(load=="LE") %>% filter(level == "HH"),aes(x=Time,ymin=SE_min,ymax=SE_max),fill="green",alpha=.2,linetype=2)+
    ylab("Mean Activity") +
    ggtitle("High Performing Subjects")+
    ylim(c(-.2,.4))+
    scale_colour_manual(values=high_col)
  
  print((low + high) +
          plot_annotation(title= names(cue_TC_for_plot)[[ROI]])+
          plot_layout(guides = "collect"))
  
}

Delay

low_col = c("turquoise","blue","violet")
high_col = c("red","gold","green")

for (ROI in seq.int(1,length(delay_TC_for_plot))){
  low <- ggplot(data=delay_TC_for_plot[[ROI]][["wide"]] %>% filter(level == "LL" | level == "LM" | level == "LH")) +
    geom_rect(data=rects,aes(xmin=xstart, xmax=xend, ymin = -Inf, ymax=Inf), fill="grey", alpha =0.4,show.legend = FALSE)+
    geom_line(aes(x=Time,y=LE,color=level),size=1)+
    geom_ribbon(data=delay_TC_for_plot[[ROI]][["long"]] %>% filter(load=="LE") %>% filter(level == "LL"),aes(x=Time,ymin=SE_min,ymax=SE_max),alpha=.2,linetype=2,fill="turquoise")+
    geom_ribbon(data=delay_TC_for_plot[[ROI]][["long"]] %>% filter(load=="LE") %>% filter(level == "LM"),aes(x=Time,ymin=SE_min,ymax=SE_max),alpha=.2,linetype=2,fill="blue")+
    geom_ribbon(data=delay_TC_for_plot[[ROI]][["long"]] %>% filter(load=="LE") %>% filter(level == "LH"),aes(x=Time,ymin=SE_min,ymax=SE_max),alpha=.2,linetype=2,fill="violet")+
    ylab("Mean Activity") +
    ggtitle("Low Performing Subjects")+
    ylim(c(-.2,.4))+
    scale_colour_manual(values=low_col)
  
  high <- ggplot(data=delay_TC_for_plot[[ROI]][["wide"]] %>% filter(level == "HL" | level == "HM" | level == "HH")) +
    geom_rect(data=rects,aes(xmin=xstart, xmax=xend, ymin = -Inf, ymax=Inf), fill="grey", alpha =0.4,show.legend = FALSE)+
    geom_line(aes(x=Time,y=LE,color=level),size=1)+
    geom_ribbon(data=delay_TC_for_plot[[ROI]][["long"]] %>% filter(load=="LE") %>% filter(level == "HL"),aes(x=Time,ymin=SE_min,ymax=SE_max),fill="red",alpha=.2,linetype=2)+
    geom_ribbon(data=delay_TC_for_plot[[ROI]][["long"]] %>% filter(load=="LE") %>% filter(level == "HM"),aes(x=Time,ymin=SE_min,ymax=SE_max),fill="gold",alpha=.2,linetype=2)+
    geom_ribbon(data=delay_TC_for_plot[[ROI]][["long"]] %>% filter(load=="LE") %>% filter(level == "HH"),aes(x=Time,ymin=SE_min,ymax=SE_max),fill="green",alpha=.2,linetype=2)+
    ylab("Mean Activity") +
    ggtitle("High Performing Subjects")+
    ylim(c(-.2,.4))+
    scale_colour_manual(values=high_col)
  
  print((low + high) +
          plot_annotation(title= names(delay_TC_for_plot)[[ROI]])+
          plot_layout(guides = "collect"))
  
}

Probe

low_col = c("turquoise","blue","violet")
high_col = c("red","gold","green")

for (ROI in seq.int(1,length(probe_TC_for_plot))){
  low <- ggplot(data=probe_TC_for_plot[[ROI]][["wide"]] %>% filter(level == "LL" | level == "LM" | level == "LH")) +
    geom_rect(data=rects,aes(xmin=xstart, xmax=xend, ymin = -Inf, ymax=Inf), fill="grey", alpha =0.4,show.legend = FALSE)+
    geom_line(aes(x=Time,y=LE,color=level),size=1)+
    geom_ribbon(data=probe_TC_for_plot[[ROI]][["long"]] %>% filter(load=="LE") %>% filter(level == "LL"),aes(x=Time,ymin=SE_min,ymax=SE_max),alpha=.2,linetype=2,fill="turquoise")+
    geom_ribbon(data=probe_TC_for_plot[[ROI]][["long"]] %>% filter(load=="LE") %>% filter(level == "LM"),aes(x=Time,ymin=SE_min,ymax=SE_max),alpha=.2,linetype=2,fill="blue")+
    geom_ribbon(data=probe_TC_for_plot[[ROI]][["long"]] %>% filter(load=="LE") %>% filter(level == "LH"),aes(x=Time,ymin=SE_min,ymax=SE_max),alpha=.2,linetype=2,fill="violet")+
    ylab("Mean Activity") +
    ggtitle("Low Performing Subjects")+
    ylim(c(-.2,.4))+
    scale_colour_manual(values=low_col)
  
  high <- ggplot(data=probe_TC_for_plot[[ROI]][["wide"]] %>% filter(level == "HL" | level == "HM" | level == "HH")) +
    geom_rect(data=rects,aes(xmin=xstart, xmax=xend, ymin = -Inf, ymax=Inf), fill="grey", alpha =0.4,show.legend = FALSE)+
    geom_line(aes(x=Time,y=LE,color=level),size=1)+
    geom_ribbon(data=probe_TC_for_plot[[ROI]][["long"]] %>% filter(load=="LE") %>% filter(level == "HL"),aes(x=Time,ymin=SE_min,ymax=SE_max),fill="red",alpha=.2,linetype=2)+
    geom_ribbon(data=probe_TC_for_plot[[ROI]][["long"]] %>% filter(load=="LE") %>% filter(level == "HM"),aes(x=Time,ymin=SE_min,ymax=SE_max),fill="gold",alpha=.2,linetype=2)+
    geom_ribbon(data=probe_TC_for_plot[[ROI]][["long"]] %>% filter(load=="LE") %>% filter(level == "HH"),aes(x=Time,ymin=SE_min,ymax=SE_max),fill="green",alpha=.2,linetype=2)+
    ylab("Mean Activity") +
    ggtitle("High Performing Subjects")+
    ylim(c(-.2,.4))+
    scale_colour_manual(values=high_col)
  
  print((low + high) +
          plot_annotation(title= names(probe_TC_for_plot)[[ROI]])+
          plot_layout(guides = "collect"))
  
}

Split on WM

Same idea, but split on WMC.

Cue

low_col <- c("red","turquoise")
med_col <- c("gold","blue")
high_col <- c("green","violet")

for (ROI in seq.int(1,length(cue_TC_for_plot))){
  low <- ggplot(data=cue_TC_for_plot[[ROI]][["wide"]] %>% filter(level == "LL" | level == "HL")) +
    geom_rect(data=rects,aes(xmin=xstart, xmax=xend, ymin = -Inf, ymax=Inf), fill="grey", alpha =0.4,show.legend = FALSE)+
    geom_line(aes(x=Time,y=LE,color=level),size=1)+
    geom_ribbon(data=cue_TC_for_plot[[ROI]][["long"]] %>% filter(load=="LE") %>% filter(level == "LL"),aes(x=Time,ymin=SE_min,ymax=SE_max),alpha=.2,linetype=2,fill="turquoise")+
    geom_ribbon(data=cue_TC_for_plot[[ROI]][["long"]] %>% filter(load=="LE") %>% filter(level == "HL"),aes(x=Time,ymin=SE_min,ymax=SE_max),alpha=.2,linetype=2,fill="red")+
    ylab("Mean Activity") +
    ggtitle("Low WMC")+
    ylim(c(-.2,.4))+
    scale_colour_manual(values=low_col)
  
  med <- ggplot(data=cue_TC_for_plot[[ROI]][["wide"]] %>% filter(level == "LM" | level == "HM")) +
    geom_rect(data=rects,aes(xmin=xstart, xmax=xend, ymin = -Inf, ymax=Inf), fill="grey", alpha =0.4,show.legend = FALSE)+
    geom_line(aes(x=Time,y=LE,color=level),size=1)+
    geom_ribbon(data=cue_TC_for_plot[[ROI]][["long"]] %>% filter(load=="LE") %>% filter(level == "LM"),aes(x=Time,ymin=SE_min,ymax=SE_max),alpha=.2,linetype=2,fill="blue")+
    geom_ribbon(data=cue_TC_for_plot[[ROI]][["long"]] %>% filter(load=="LE") %>% filter(level == "HM"),aes(x=Time,ymin=SE_min,ymax=SE_max),alpha=.2,linetype=2,fill="gold")+
    ylab("Mean Activity") +
    ggtitle("Med WMC")+
    ylim(c(-.2,.4))+
    scale_colour_manual(values=med_col)
  
  high <- ggplot(data=cue_TC_for_plot[[ROI]][["wide"]] %>% filter(level == "LH" | level == "HH")) +
    geom_rect(data=rects,aes(xmin=xstart, xmax=xend, ymin = -Inf, ymax=Inf), fill="grey", alpha =0.4,show.legend = FALSE)+
    geom_line(aes(x=Time,y=LE,color=level),size=1)+
    geom_ribbon(data=cue_TC_for_plot[[ROI]][["long"]] %>% filter(load=="LE") %>% filter(level == "LH"),aes(x=Time,ymin=SE_min,ymax=SE_max),alpha=.2,linetype=2,fill="violet")+
    geom_ribbon(data=cue_TC_for_plot[[ROI]][["long"]] %>% filter(load=="LE") %>% filter(level == "HH"),aes(x=Time,ymin=SE_min,ymax=SE_max),alpha=.2,linetype=2,fill="green")+
    ylab("Mean Activity") +
    ggtitle("High WMC")+
    ylim(c(-.2,.4))+
    scale_colour_manual(values=high_col)
  
  print((low + med + high) +
          plot_annotation(title= names(cue_TC_for_plot)[[ROI]])+
          plot_layout(guides = "collect")
  )
  
}

Delay

low_col <- c("red","turquoise")
med_col <- c("gold","blue")
high_col <- c("green","violet")

for (ROI in seq.int(1,length(delay_TC_for_plot))){
  low <- ggplot(data=delay_TC_for_plot[[ROI]][["wide"]] %>% filter(level == "LL" | level == "HL")) +
    geom_rect(data=rects,aes(xmin=xstart, xmax=xend, ymin = -Inf, ymax=Inf), fill="grey", alpha =0.4,show.legend = FALSE)+
    geom_line(aes(x=Time,y=LE,color=level),size=1)+
    geom_ribbon(data=delay_TC_for_plot[[ROI]][["long"]] %>% filter(load=="LE") %>% filter(level == "LL"),aes(x=Time,ymin=SE_min,ymax=SE_max),alpha=.2,linetype=2,fill="turquoise")+
    geom_ribbon(data=delay_TC_for_plot[[ROI]][["long"]] %>% filter(load=="LE") %>% filter(level == "HL"),aes(x=Time,ymin=SE_min,ymax=SE_max),alpha=.2,linetype=2,fill="red")+
    ylab("Mean Activity") +
    ggtitle("Low WMC")+
    ylim(c(-.2,.4))+
    scale_colour_manual(values=low_col)
  
  med <- ggplot(data=delay_TC_for_plot[[ROI]][["wide"]] %>% filter(level == "LM" | level == "HM")) +
    geom_rect(data=rects,aes(xmin=xstart, xmax=xend, ymin = -Inf, ymax=Inf), fill="grey", alpha =0.4,show.legend = FALSE)+
    geom_line(aes(x=Time,y=LE,color=level),size=1)+
    geom_ribbon(data=delay_TC_for_plot[[ROI]][["long"]] %>% filter(load=="LE") %>% filter(level == "LM"),aes(x=Time,ymin=SE_min,ymax=SE_max),alpha=.2,linetype=2,fill="blue")+
    geom_ribbon(data=delay_TC_for_plot[[ROI]][["long"]] %>% filter(load=="LE") %>% filter(level == "HM"),aes(x=Time,ymin=SE_min,ymax=SE_max),alpha=.2,linetype=2,fill="gold")+
    ylab("Mean Activity") +
    ggtitle("Med WMC")+
    ylim(c(-.2,.4))+
    scale_colour_manual(values=med_col)
  
  high <- ggplot(data=delay_TC_for_plot[[ROI]][["wide"]] %>% filter(level == "LH" | level == "HH")) +
    geom_rect(data=rects,aes(xmin=xstart, xmax=xend, ymin = -Inf, ymax=Inf), fill="grey", alpha =0.4,show.legend = FALSE)+
    geom_line(aes(x=Time,y=LE,color=level),size=1)+
    geom_ribbon(data=delay_TC_for_plot[[ROI]][["long"]] %>% filter(load=="LE") %>% filter(level == "LH"),aes(x=Time,ymin=SE_min,ymax=SE_max),alpha=.2,linetype=2,fill="violet")+
    geom_ribbon(data=delay_TC_for_plot[[ROI]][["long"]] %>% filter(load=="LE") %>% filter(level == "HH"),aes(x=Time,ymin=SE_min,ymax=SE_max),alpha=.2,linetype=2,fill="green")+
    ylab("Mean Activity") +
    ggtitle("High WMC")+
    ylim(c(-.2,.4))+
    scale_colour_manual(values=high_col)
  
  print((low + med + high) +
          plot_annotation(title= names(delay_TC_for_plot)[[ROI]])+
          plot_layout(guides = "collect"))
  
}

Probe

low_col <- c("red","turquoise")
med_col <- c("gold","blue")
high_col <- c("green","violet")

for (ROI in seq.int(1,length(probe_TC_for_plot))){
  low <- ggplot(data=probe_TC_for_plot[[ROI]][["wide"]] %>% filter(level == "LL" | level == "HL")) +
    geom_rect(data=rects,aes(xmin=xstart, xmax=xend, ymin = -Inf, ymax=Inf), fill="grey", alpha =0.4,show.legend = FALSE)+
    geom_line(aes(x=Time,y=LE,color=level),size=1)+
    geom_ribbon(data=probe_TC_for_plot[[ROI]][["long"]] %>% filter(load=="LE") %>% filter(level == "LL"),aes(x=Time,ymin=SE_min,ymax=SE_max),alpha=.2,linetype=2,fill="turquoise")+
    geom_ribbon(data=probe_TC_for_plot[[ROI]][["long"]] %>% filter(load=="LE") %>% filter(level == "HL"),aes(x=Time,ymin=SE_min,ymax=SE_max),alpha=.2,linetype=2,fill="red")+
    ylab("Mean Activity") +
    ggtitle("Low WMC")+
    ylim(c(-.2,.4))+
    scale_colour_manual(values=low_col)
  
  med <- ggplot(data=probe_TC_for_plot[[ROI]][["wide"]] %>% filter(level == "LM" | level == "HM")) +
    geom_rect(data=rects,aes(xmin=xstart, xmax=xend, ymin = -Inf, ymax=Inf), fill="grey", alpha =0.4,show.legend = FALSE)+
    geom_line(aes(x=Time,y=LE,color=level),size=1)+
    geom_ribbon(data=probe_TC_for_plot[[ROI]][["long"]] %>% filter(load=="LE") %>% filter(level == "LM"),aes(x=Time,ymin=SE_min,ymax=SE_max),alpha=.2,linetype=2,fill="blue")+
    geom_ribbon(data=probe_TC_for_plot[[ROI]][["long"]] %>% filter(load=="LE") %>% filter(level == "HM"),aes(x=Time,ymin=SE_min,ymax=SE_max),alpha=.2,linetype=2,fill="gold")+
    ylab("Mean Activity") +
    ggtitle("Med WMC")+
    ylim(c(-.2,.4))+
    scale_colour_manual(values=med_col)
  
  high <- ggplot(data=probe_TC_for_plot[[ROI]][["wide"]] %>% filter(level == "LH" | level == "HH")) +
    geom_rect(data=rects,aes(xmin=xstart, xmax=xend, ymin = -Inf, ymax=Inf), fill="grey", alpha =0.4,show.legend = FALSE)+
    geom_line(aes(x=Time,y=LE,color=level),size=1)+
    geom_ribbon(data=probe_TC_for_plot[[ROI]][["long"]] %>% filter(load=="LE") %>% filter(level == "LH"),aes(x=Time,ymin=SE_min,ymax=SE_max),alpha=.2,linetype=2,fill="violet")+
    geom_ribbon(data=probe_TC_for_plot[[ROI]][["long"]] %>% filter(load=="LE") %>% filter(level == "HH"),aes(x=Time,ymin=SE_min,ymax=SE_max),alpha=.2,linetype=2,fill="green")+
    ylab("Mean Activity") +
    ggtitle("High WMC")+
    ylim(c(-.2,.4))+
    scale_colour_manual(values=high_col)
  
  print((low + med + high) +
          plot_annotation(title= names(probe_TC_for_plot)[[ROI]])+
          plot_layout(guides = "collect"))
  
}

WMC and DFR Split Groups Comparison

Catherine Walsh

4/8/2020

Regressions

Cue Period Load Effect

Delay Period

High Load

Load Effect

Probe Period LE

FFA

Time courses

L1 vs L3 activity

Cue

Delay

Probe

Load Effects

Cue

Delay

Probe

Split on DFR

Cue

Delay

Probe

Split on WM

Cue

Delay

Probe